“Data is Destiny:” The Importance of Data Quality in the Age of Artificial Intelligence
There was a time where data collection was onerous and assessing the data was manual. This meant that organizations had to strike a critical balance between the speed and quality of their decision-making. We’ve always made decisions based on data; the process was slow, labor intensive, and required focus on a limited set of variables. There can be no denying that technology has changed the speed and scale of DDDM. With the capabilities present today in elastic cloud computing, big data, AI, and the Internet of Things, our ability to collect, process, and analyze data to make decisions has become exponentially faster and more wide-reaching. This automatically means better outcomes, right?
Consider the words of Joy Buolamwini, the MIT student who uncovered algorithmic bias in facial recognition programs and is a central figure in the Netflix documentary, Coded Bias. Buolamwini says, “If you are thinking about data and artificial intelligence, in many ways data is destiny. It is what we are using to teach machines different kinds of patterns. If you have largely skewed datasets that are being used to train these systems, you can also have skewed results.” She said this in the context of racial bias in facial recognition programs, but the exact same principle can be applied to any AI model in a business or industrial setting.
Intelligent is not the default state of an AI model. AI models need to be taught how to think and they are taught by the data they are fed. An AI algorithm literally knows nothing else of the world but the 0s and 1s it is provided. In this context, you see the immense importance that is placed on data quantity and quality. As an organization digitally transforms its operations, data most certainly is destiny. Decisions in a digital world will always be faster, but whether they are better has a direct dependence on the quality of the input data.
Let me be clear, I do not say these things to diminish the promise of digital transformation or AI. On the contrary, I believe that organizations that do not embrace digital transformation will eventually become extinct. Slow never wins the race unless ALL other participants make crucial mistakes. I do, however, think it is critical to highlight the importance of addressing data quality as a foundational element to digital transformation.
Any digital transformation program should be built on a foundation of, at least, the following:
- Data standards optimized for data interoperability
- Data governance to ensure adherence to standards
- Strategic IT infrastructure roadmap with data interoperability as a primary objective